You can try lowering the number of jacobi iterations, the fluid resolution, the resolution of the rendering (one ray is shot per pixel), the number of steps for the shadows which are expensive the way it’s done…
But the geforce 285 is quite older than the geforce 4XX no ? So it doesn’t seem incoherent. Fluid simulation is always very slow, it’s even worse in software…
True. I still haven’t had the time to check out the code but I’m sure there are alot of params to tweak. Are you using any space skipping algorithm to skip over big empty areas of the volume? seems like it would help in the case of a shaft of smoke inside a cubic volume.